The Department's Committee on Instruction would like to suggest an area of potential instructional improvement in our deparment, namely, providing predictive grade feedback to students throughout the quarter.D.1
The current status is that many courses do inform students of raw scores achieved during the quarter. Some courses also provide statistics and sorted lists. However, we have found that most courses do not provide adequate predictive grade feedback to the students. Grade feedback means to inform the students what letter grade they are receiving throughout the quarter. Predictive grade feedback means that the letter grade information received during the quarter has a defined contribution to the final course grade. As a simple example, if mid-quarter feedback is given after 50% of a course's total points have been graded, and a student received a middle C on that 50%, then that student would need a middle A on the remaining 50% to received a middle B.
The current approach to grading seems to be that raw scores are combined as a weighted sum (based on the syllabus' weighting of components) at the end of the quarter, students are sorted by that weighted sum, and lines are drawn by the instructor. The problem with this approach is that students do not really know how they are doing in the course during the quarter. In talking with students, we have found that the vast majority believe they are doing much better than they really are. Thus, they do not pick up their pace appropriately, or drop/withdraw when they really should have. Without predictive grade data during the quarter, they tend to hope the instructor will be generous. A very common complaint among students, including our best students, is that they thought they were getting an A (or B or whatever) and then ended up with a B (or C or whatever), and they don't understand why; likewise, students often say that they would have dropped a course had they known how they were really doing in the instructor's eye.
Simply showing how students are doing relative to the average is not sufficient, as students do not know what the average represents: Is it a C, B, D, or even an F, as is the case in some courses?
Exacerbating the problem is that TAs and graders do not know how to grade assignments in a way that scores can be meaningfully combined together. One grader may create a scoring scheme on one assignement where 90% and up represents A-quality work, whereas another may create a scheme on a different assignment where 80% and up represents an A. Combining raw scores for those two assignments results in the first assignment unintentionally carry much more weight than the second.
A simple way (but not the only way) to provide predictive grading is to use what is generally known as a conventional 90/80/70/60 scale throughout the quarter. In other words, all material (homeworks, quizzes, exams, programs, etc.) are created and graded such that receiving 90% of the points on that assignment represents an A, 80% a B, 70% a C, and 60% a D.
If an instructor finds that a graded item (typically an exam) doesn't fall into the 90/80/70/60 model, then that item can be scaled to a 90/80/70/60. As a simple example, if the average on a 100-point exam turns out to be a 60 and the high scores around 90, and if the instructor believes that a 60 low C quality work, then the instructor might simply scale by adding 10 points to every exam, reducing the total points possible to 90, or multiplying every score by 100/90. There are many other ways to scale.D.2 In any case, the student is given his raw score and scaled score when the graded item is returned. Thus, the student knows whether he's doing A work, B work, etc., simply by seeing whether the scale score is 90%, 80%, etc.
By using a 90/80/70/60 scale and scaling items throughout the quarter, the cumulative course percentage also represents the student's total course grade, and thus posted grades throughout the quarter have a predictive feature. For example, if a student has received 85% on the first 50% of course points so far, he needs a 95% on the remaining material to get an A (90%).D.3 If instead a student has received 40% on the first 50%, he would need 100% on the remaining points to get a C- (70%) -- given such information, that student would likely withdraw if possible.
Note that the above is not the same as curving. Curving usually means to give a certain percentage of students As, a certain percentage Bs, etc. The above says nothing about what defines A-quality work, B-quality work, etc. It simply says that, once the instructor has defined that for a given item, the instructor should scale the score to a 90/80/70/60 point system, in order to provide predictive feedback.
An instructor may still want to adjust the final grade cutoffs to assist the students if the instructors feels the grades are too harsh, e.g., adjust the final scale to 89/78/67/56. Of course, if huge adjustments are made (e.g., to a 80/60/40/20 scale), then the predictive nature of the grades throughout the quarter is greatly diminished. It would be better to try to create and scale items throughout the quarter to stay as close to the 90/80/70/60 scale as possible. Also, we probably should not adjust final cutoffs in the opposite direction (e.g., 91/82/73/64), as the predictive grade quality of most interest to the student is the minimum grade they'll receive.
Implicit throughout the discussion above is that instructors create assessment items throughout the quarter, and not leave all the graded work until the end of the quarter. Instructors should strive to ensure that a good percentage of course points have been reported to the students well before the last day to withdraw (end of sixth week). Preferably, some feedback would be given before the last day to drop (third week).
There are many other ways to provide predictive grade feedback, but the above seems to be fairly common in many universities and easy to understand by TAs and students. The above will become policy for the CS10/12/14/61 courses, and thus our students will be very familiar with the system when they enter the upper division. We would do our students a good service by providing such feedback more consistently throughout our courses, moving the grade-assignment process from a backroom, end-of-the-quarter, hidden process, to a more transparent process that students can use to realize when more effort is needed.